.. _sec_sentiment_rnn:
Análise de Sentimento: Usando Redes Neurais Recorrentes
=======================================================
Semelhante a sinônimos e analogias de pesquisa, a classificação de texto
também é uma aplicação posterior de incorporação de palavras. Nesta
seção, vamos aplicar vetores de palavras pré-treinados (GloVe) e redes
neurais recorrentes bidirecionais com múltiplas camadas ocultas
:cite:`Maas.Daly.Pham.ea.2011`, como mostrado em
:numref:`fig_nlp-map-sa-rnn`. Usaremos o modelo para determinar se uma
sequência de texto de comprimento indefinido contém emoção positiva ou
negativa.
.. _fig_nlp-map-sa-rnn:
.. figure:: ../img/nlp-map-sa-rnn.svg
Esta seção alimenta o GloVe pré-treinado para uma arquitetura baseada
em RNN para análise de sentimento.
.. raw:: html
.. raw:: html
.. code:: python
from mxnet import gluon, init, np, npx
from mxnet.gluon import nn, rnn
from d2l import mxnet as d2l
npx.set_np()
batch_size = 64
train_iter, test_iter, vocab = d2l.load_data_imdb(batch_size)
.. raw:: html
.. raw:: html
.. code:: python
import torch
from torch import nn
from d2l import torch as d2l
batch_size = 64
train_iter, test_iter, vocab = d2l.load_data_imdb(batch_size)
.. parsed-literal::
:class: output
Downloading ../data/aclImdb_v1.tar.gz from http://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz...
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
class BiRNN(nn.Block):
def __init__(self, vocab_size, embed_size, num_hiddens,
num_layers, **kwargs):
super(BiRNN, self).__init__(**kwargs)
self.embedding = nn.Embedding(vocab_size, embed_size)
# Set `bidirectional` to True to get a bidirectional recurrent neural
# network
self.encoder = rnn.LSTM(num_hiddens, num_layers=num_layers,
bidirectional=True, input_size=embed_size)
self.decoder = nn.Dense(2)
def forward(self, inputs):
# The shape of `inputs` is (batch size, no. of words). Because LSTM
# needs to use sequence as the first dimension, the input is
# transformed and the word feature is then extracted. The output shape
# is (no. of words, batch size, word vector dimension).
embeddings = self.embedding(inputs.T)
# Since the input (embeddings) is the only argument passed into
# rnn.LSTM, it only returns the hidden states of the last hidden layer
# at different time step (outputs). The shape of `outputs` is
# (no. of words, batch size, 2 * no. of hidden units).
outputs = self.encoder(embeddings)
# Concatenate the hidden states of the initial time step and final
# time step to use as the input of the fully connected layer. Its
# shape is (batch size, 4 * no. of hidden units)
encoding = np.concatenate((outputs[0], outputs[-1]), axis=1)
outs = self.decoder(encoding)
return outs
.. raw:: html
.. raw:: html
.. code:: python
class BiRNN(nn.Module):
def __init__(self, vocab_size, embed_size, num_hiddens,
num_layers, **kwargs):
super(BiRNN, self).__init__(**kwargs)
self.embedding = nn.Embedding(vocab_size, embed_size)
# Set `bidirectional` to True to get a bidirectional recurrent neural
# network
self.encoder = nn.LSTM(embed_size, num_hiddens, num_layers=num_layers,
bidirectional=True)
self.decoder = nn.Linear(4*num_hiddens, 2)
def forward(self, inputs):
# The shape of `inputs` is (batch size, no. of words). Because LSTM
# needs to use sequence as the first dimension, the input is
# transformed and the word feature is then extracted. The output shape
# is (no. of words, batch size, word vector dimension).
embeddings = self.embedding(inputs.T)
# Since the input (embeddings) is the only argument passed into
# nn.LSTM, both h_0 and c_0 default to zero.
# we only use the hidden states of the last hidden layer
# at different time step (outputs). The shape of `outputs` is
# (no. of words, batch size, 2 * no. of hidden units).
self.encoder.flatten_parameters()
outputs, _ = self.encoder(embeddings)
# Concatenate the hidden states of the initial time step and final
# time step to use as the input of the fully connected layer. Its
# shape is (batch size, 4 * no. of hidden units)
encoding = torch.cat((outputs[0], outputs[-1]), dim=1)
outs = self.decoder(encoding)
return outs
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
embed_size, num_hiddens, num_layers, devices = 100, 100, 2, d2l.try_all_gpus()
net = BiRNN(len(vocab), embed_size, num_hiddens, num_layers)
net.initialize(init.Xavier(), ctx=devices)
.. raw:: html
.. raw:: html
.. code:: python
embed_size, num_hiddens, num_layers, devices = 100, 100, 2, d2l.try_all_gpus()
net = BiRNN(len(vocab), embed_size, num_hiddens, num_layers)
def init_weights(m):
if type(m) == nn.Linear:
nn.init.xavier_uniform_(m.weight)
if type(m) == nn.LSTM:
for param in m._flat_weights_names:
if "weight" in param:
nn.init.xavier_uniform_(m._parameters[param])
net.apply(init_weights);
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
glove_embedding = d2l.TokenEmbedding('glove.6b.100d')
.. raw:: html
.. raw:: html
.. code:: python
glove_embedding = d2l.TokenEmbedding('glove.6b.100d')
.. parsed-literal::
:class: output
Downloading ../data/glove.6B.100d.zip from http://d2l-data.s3-accelerate.amazonaws.com/glove.6B.100d.zip...
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
embeds = glove_embedding[vocab.idx_to_token]
embeds.shape
.. parsed-literal::
:class: output
(49346, 100)
.. raw:: html
.. raw:: html
.. code:: python
embeds = glove_embedding[vocab.idx_to_token]
embeds.shape
.. parsed-literal::
:class: output
torch.Size([49346, 100])
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
net.embedding.weight.set_data(embeds)
net.embedding.collect_params().setattr('grad_req', 'null')
.. raw:: html
.. raw:: html
.. code:: python
net.embedding.weight.data.copy_(embeds)
net.embedding.weight.requires_grad = False
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
lr, num_epochs = 0.01, 5
trainer = gluon.Trainer(net.collect_params(), 'adam', {'learning_rate': lr})
loss = gluon.loss.SoftmaxCrossEntropyLoss()
d2l.train_ch13(net, train_iter, test_iter, loss, trainer, num_epochs, devices)
.. parsed-literal::
:class: output
loss 0.323, train acc 0.865, test acc 0.842
444.5 examples/sec on [gpu(0), gpu(1)]
.. figure:: output_sentiment-analysis-rnn_6199ad_57_1.svg
.. raw:: html
.. raw:: html
.. code:: python
lr, num_epochs = 0.01, 5
trainer = torch.optim.Adam(net.parameters(), lr=lr)
loss = nn.CrossEntropyLoss(reduction="none")
d2l.train_ch13(net, train_iter, test_iter, loss, trainer, num_epochs, devices)
.. parsed-literal::
:class: output
loss 0.333, train acc 0.860, test acc 0.829
749.0 examples/sec on [device(type='cuda', index=0), device(type='cuda', index=1)]
.. figure:: output_sentiment-analysis-rnn_6199ad_60_1.svg
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
#@save
def predict_sentiment(net, vocab, sentence):
sentence = np.array(vocab[sentence.split()], ctx=d2l.try_gpu())
label = np.argmax(net(sentence.reshape(1, -1)), axis=1)
return 'positive' if label == 1 else 'negative'
.. raw:: html
.. raw:: html
.. code:: python
#@save
def predict_sentiment(net, vocab, sentence):
sentence = torch.tensor(vocab[sentence.split()], device=d2l.try_gpu())
label = torch.argmax(net(sentence.reshape(1, -1)), dim=1)
return 'positive' if label == 1 else 'negative'
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
predict_sentiment(net, vocab, 'this movie is so great')
.. parsed-literal::
:class: output
'positive'
.. code:: python
predict_sentiment(net, vocab, 'this movie is so bad')
.. parsed-literal::
:class: output
'negative'
.. raw:: html
.. raw:: html
.. code:: python
predict_sentiment(net, vocab, 'this movie is so great')
.. parsed-literal::
:class: output
'positive'
.. code:: python
predict_sentiment(net, vocab, 'this movie is so bad')
.. parsed-literal::
:class: output
'negative'
.. raw:: html
.. raw:: html
.. raw:: html
`Discussões `__
.. raw:: html
.. raw:: html
`Discussões `__
.. raw:: html
.. raw:: html